Open-ended Generative VQA on IRIS-120K Tree-VQA
42.36BLEU-1IRIS-4B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| IRIS-4BGroup=Ours2026.07 | 42.36 | 39.29 | 29.03 | |
| IRIS-8BGroup=Ours2026.07 | 42.27 | 39.68 | 29.64 | |
| IRIS-4B-TFTGroup=Ours2026.07 | 41.86 | 39.18 | 28.95 | |
| IRIS-2BGroup=Ours2026.07 | 40.4 | 38.06 | 27.77 | |
| IRIS-4B-SceneGroup=Ours2026.07 | 37.02 | 34.31 | 24.48 | |
| HuluMed-32BGroup=Med2026.07 | 33.02 | 31.06 | 21.11 | |
| HuluMed-14BGroup=Med2026.07 | 32.48 | 30.68 | 20.8 | |
| HuluMed-7BGroup=Med2026.07 | 31.91 | 30.17 | 20.58 | |
| HGPT-V-34BGroup=Med2026.07 | 28.6 | 27.18 | 18.79 | |
| Lingshu-32BGroup=Med2026.07 | 28.04 | 29.85 | 21.92 | |
| Lingshu-7BGroup=Med2026.07 | 21.63 | 25.88 | 19.33 | |
| HGPT-V-7BGroup=Med2026.07 | 21.34 | 22.65 | 15.02 | |
| GPT-5-miniGroup=Gen2026.07 | 16.48 | 11.31 | 21.81 | |
| MG-4BGroup=Med2026.07 | 16.13 | 22.68 | 16.52 | |
| Qwen3-vl-32BGroup=Gen2026.07 | 13.65 | 20.79 | 13.06 | |
| Internvl3.5-14BGroup=Gen2026.07 | 11.82 | 19.8 | 11.59 | |
| MG1.5-4BGroup=Med2026.07 | 11.52 | 17.19 | 9.98 | |
| Internvl3.5-8BGroup=Gen2026.07 | 10.52 | 17.94 | 10.64 | |
| Qwen3-vl-8BGroup=Gen2026.07 | 7.77 | 16.3 | 8.88 | |
| Qwen3-vl-4BGroup=Gen2026.07 | 6.19 | 14.24 | 7.83 | |
| Qwen3-vl-2BGroup=Gen2026.07 | 4.82 | 15.88 | 9.06 |